DVCon 2009 Wrap Up: Attendance
Last Minute Deal on SystemVerilog Training with Cliff Cummings in Beaverton, Oregon, 4/21-24

SNUG Plug: Multi-Stream Scenarios and the VMM

Ok, I know what you’re all thinking… “JL, you haven’t even finished writing up DVCon yet and now you’re talking about SNUG?!” Yes, I know.  Let’s just say an annoying stream of illnesses have taken their toll on the Gray family (and many other folks as best as I can tell) over the last few weeks.  But things seem almost back to normal now and I didn’t want to miss out on an opportunity to let everyone know about my official debut as a conference paper presenter at SNUG San Jose next week. 

My Verilab colleagues Jason Sprott and Sumit Dhamanwala, along with Cliff Cummings from Sunburst Design and yours truly authored a paper entitled “Using the New Features in VMM 1.1 for Multi-Stream Scenarios”. I’ll be presenting the paper during session MA4: Verification with VMM I on Monday at 11am.  Those of you who attended one of the Verification Now 2008 seminars back in the fall will recognize the topic. I discussed the yet to be announced Multi-Stream Scenario additions to the VMM in one of my presentations. 

Unlike my Verification Now presentation which compared stimulus in the OVM to the VMM, the SNUG presentation will delve into the topic of Multi-Stream Scenarios in the VMM in more detail.  Specifically, I will review the following topics:

  • Recap: Single Stream Scenarios
  • Complex Stimulus with Multi-Stream Scenarios
  • Multi-Stream Scenario Registries (Channel, MSS, and MSSG)
  • Single Stream vs. Multi-Stream Scenarios
  • Resource Sharing: Grab/Ungrab
  • Multi-channel grab

One of the things I hope to touch on is the importance of using the registries when building multi-stream scenarios instead of directly instantiating sub-scenarios, channels, or scenarios from other multi-stream scenario generators (using the generator registry).  Those features were added to the MSS solution to allow integrators and test writers to modify the behavior of specific scenarios and scenario generators without having to modify the underlying scenarios themselves. 

Another goal is simply to promote the topic of reusable, multi-stream stimulus itself. The VMM has historically supported a flat testbench structure. New features such as MSS, when combined with the vmm_subenv should lead to more reusable and maintainable testbenches.

I’m thrilled to have an opportunity to present at SNUG this year and to meet readers of this blog and my Twitter feed.  I will be twittering SNUG using the hash tag #snug, unless someone gives me a good reason to use a different tag.  Of course, that begs the question – who will twitter my presentation while I’m presenting?  How about this… I will buy the person who Twitters the most insightful comments and/or questions during my presentation a delicious beverage of his/her choice. Any takers?